What the Data for This Category Looks Like
Data sources for this category include the MES system, equipment FDC module, defect detection and metrology equipment, wafer map files, process recipe library, and experimental data. Daily documents cover process flow cards, batch operation records, equipment recipes, SPC charts, exception handling sheets, defect maps, and yield reports. Data updates occur after each wafer batch completes, via real-time parameter collection, and daily summary updates. Fields include wafer batch identifiers, nanoscale process parameters, equipment runtime, yield values, and more. Units include nanometers, hours, percentages, and others.
What Constraints These Characteristics Impose on Deployment and Upgrade
Multi-source heterogeneous data sources require permission and format compatibility configurations for multi-system integration during deployment. The high-frequency update rhythm creates performance pressure for real-time data synchronization and batch ledger summary, requiring targeted adjustments to data caching and batch processing parameters. Diverse unstructured document types (wafer maps, SPC charts, defect maps) require configuration of corresponding format parsing rules. The high-precision requirement for nanoscale process parameters requires ensuring that data storage and retrieval precision align with business requirements. Upgrades must retain format compatibility for existing data.
How to Configure the Settings
The following configurations apply to FastGPT v4.17.0.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
DB_MAX_LINK | 20–30 | Adapts to concurrent connection requirements for multi-source data synchronization, prevents connection pool exhaustion |
MONGODB_URI | Enter the MongoDB cluster connection address for wafer ledger data, including authentication information | Stores massive structured and unstructured ledger data such as batch operation records and defect maps |
PG_URL | Enter the PostgreSQL database connection address associated with the MES system | Stores structured ledger data such as process flow cards and yield reports |
REDIS_URL | Configure to the locally deployed Redis instance address, set memory limit to 8 GB | Caches real-time collected equipment FDC parameters and wafer batch status, reduces multi-source data query latency |
STORAGE_S3_ENDPOINT | Enter the object storage endpoint address compatible with the cleanroom internal network | Stores large-volume documents such as wafer maps and SPC charts, meets internal network data storage requirements |
FILE_TOKEN_KEY | Generate a 32-bit random string as the secret key | Protects access permissions for sensitive documents such as uploaded process recipes and exception handling sheets |
Three Common Misconfigurations
- Symptom: After deploying via image packaging, attachments such as wafer maps and SPC charts cannot be parsed, and no error prompt appears in the interface. Cause:
STORAGE_S3_ENDPOINTwas not correctly configured to point to the internal network-compatible storage service during deployment, causing the system to fail to read uploaded attachments. - Symptom: AxiosError 404 status code is returned when calling the code plugin. Cause: The correct internal network access route was not configured inside the Docker container, or database connection addresses such as
PG_URLandMONGODB_URIpoint to internal network addresses that cannot be accessed from outside the container. - Symptom: After the advanced orchestration's judge node, subsequent AI conversation nodes cannot obtain the initial user question. Cause: Context passthrough configuration was not enabled during deployment, or the variable mapping for the initial user question was not correctly bound in the orchestration flow.
How to Verify Successful Configuration
- Log in to the FastGPT backend, navigate to the data connection management page, confirm that the connection status of configuration items such as
MONGODB_URI,PG_URL, andREDIS_URLshows normal. - Upload a wafer map or SPC chart document, trigger the ledger summary task, and check for records of successful format parsing in the task logs.
- Enter the Docker container, execute an internal network ping command to verify connectivity between the storage service and the database, and confirm that the configured
STORAGE_S3_ENDPOINTis accessible. - Run the advanced orchestration test flow, and verify that the initial user question can be passed to subsequent AI conversation nodes.
The values provided are common starting points and should be measured against the reader's own samples.